aibrix
aibrix copied to clipboard
Add feature flag to enable heterogenous features
🚀 Feature Description and Motivation
Right now heterogenous feature such as optimizer or request tracing is enabled by default. Enabling by default has two issues 1) It add small performance penalty 2) Requires to enable include_usage for streaming.
Goal here is to add a feature flag which can fix both issues. Along with feature flag add support for request tracing for streaming.
Use Case
GPU Optimizer
Proposed Solution
No response
cc @zhangjyr
This is completed.